Cross-Clade Ultrasensitive PCR-Based Assays To Measure HIV Persistence in Large-Cohort Studies

ABSTRACT

A small pool of infected cells persists in HIV-infected individuals receiving antiretroviral therapy (ART). Here, we developed ultrasensitive assays to precisely measure the frequency of cells harboring total HIV DNA, integrated HIV DNA, and two long terminal repeat (2-LTR) circles. These assays are performed on cell lysates, which circumvents the labor-intensive step of DNA extraction, and rely on the coquantification of each HIV molecular form together with CD3 gene sequences to precisely measure cell input. Using primary isolates from HIV subtypes A, B, C, D, and CRF01_A/E, we demonstrate that these assays can efficiently quantify low target copy numbers from diverse HIV subtypes. We further used these assays to measure total HIV DNA, integrated HIV DNA, and 2-LTR circles in CD4⁺ T cells from HIV-infected subjects infected with subtype B. All samples obtained from ART-naive subjects were positive for the three HIV molecular forms (n = 15). Total HIV DNA, integrated HIV DNA, and 2-LTR circles were detected in, respectively, 100%, 94%, and 77% of the samples from individuals in which HIV was suppressed by ART. Higher levels of total HIV DNA and 2-LTR circles were detected in untreated subjects than individuals on ART (P = 0.0003 and P = 0.0004, respectively), while the frequency of CD4⁺ T cells harboring integrated HIV DNA did not differ between the two groups. These results demonstrate that these novel assays have the ability to quantify very low levels of HIV DNA of multiple HIV subtypes without the need for nucleic acid extraction, making them well suited for the monitoring of viral persistence in large populations of HIV-infected individuals.

IMPORTANCE Since the discovery of viral reservoirs in HIV-infected subjects receiving suppressive ART, measuring the degree of viral persistence has been one of the greatest challenges in the field of HIV research. Here, we report the development and validation of ultrasensitive assays to measure HIV persistence in HIV-infected individuals from multiple geographical regions. These assays are relatively inexpensive, do not require DNA extraction, and can be completed in a single day. Therefore, they are perfectly adapted to monitor HIV persistence in large cohorts of HIV-infected individuals and, given their sensitivity, can be used to monitor the efficacy of therapeutic strategies aimed at interfering with HIV persistence after prolonged ART.

INTRODUCTION

Advances in the treatment of HIV infection have dramatically reduced the death rate from AIDS and improved the quality of life of many HIV-infected patients (1). The initiation of antiretroviral therapy (ART) results in a rapid drop in plasma viral load and in a substantial reduction in the number of cells carrying viral DNA in both blood and tissues (2). However, it is now clear that ART does not eradicate HIV, as the virus persists in long-lived latently infected CD4 T cells that can produce infectious particles upon stimulation (3–5). Two non-mutually exclusive mechanisms underlie HIV persistence in subjects who have received suppressive ART for extended periods of time. First, the incomplete suppression of viral replication by ART could allow the continuous replenishment of a small pool of infected cells (6). Low levels of persistent viremia (<50 copies/ml) are detected in a substantial fraction of so-called virally suppressed subjects (7). Increased levels of two long terminal repeat (2-LTR) circles (8, 9) as well as a significant reduction in the amount of cell-associated RNA in the gut (10) of ART subjects upon intensification with raltegravir suggest that complete cycles of viral replication may still occur during ART. The continuous replenishment of the HIV reservoir is also supported by a recent study demonstrating that lower antiretroviral drug concentrations in lymphatic tissues allow ongoing viral replication in anatomical reservoirs (11). Second, in addition to these persistent low levels of viral production (or replication) that occur in at least a subset of individuals, HIV persists in cellular reservoirs as a silent provirus. Although several cell subsets may contribute to HIV persistence (including naive CD4⁺ T cells [12] and myeloid cells [13]), the vast majority of proviral DNA is detected in CD4⁺ T cells which display a memory phenotype (14, 15). This cellular reservoir decays very slowly, with a half-life of 40 to 44 months, indicating than more than 70 years of intensive therapy would be required for its eradication (16).

An increasing number of therapeutic strategies are being tested for their ability to reduce the size of the latent HIV reservoir (17). There is, as such, an urgent need to develop a robust and precise assay scalable to a clinical trial that measures the frequency of infected cells carrying inducible replication-competent HIV. Several assays have been developed to measure the frequency of infected cells that persist in virally suppressed subjects (18). However, these assays are often low throughput and relatively expensive, which render them less feasible for clinical applications in large-cohort studies. For example, the quantitative viral outgrowth assay (19), which is often considered the “gold standard” method to measure HIV persistence during ART (20), is expensive and cumbersome, lacks precision, and does not capture all latently infected cells (21). Functionally, none of these assays reliably predicts whether patients in whom HIV is suppressed by ART will rebound during analytical treatment interruption (22). Monitoring of HIV persistence by measuring the frequency of cells harboring HIV DNA (23–25, 45, 46) represents an attractive alternative but has often been criticized, as a large fraction of the viral genomes may be replication defective (26). Notwithstanding this limitation, measuring viral DNA and, in particular, integrated HIV DNA (27, 28) has provided crucial information that has contributed to the understanding of the mechanisms of HIV persistence (14, 15, 25, 29–32). Importantly, the amount of HIV DNA has been shown to predict viral rebound and the viral set point after structured treatment interruption (33, 34). Eriksson et al. recently reported that the frequency of cells harboring integrated HIV DNA is the only measurement of the reservoir that correlates with the results of the quantitative viral outgrowth assay (20). Taken together, these observations suggest that in spite of the aforementioned limitations, the measurement of HIV DNA may provide a robust and clinically relevant measurement of the size of the HIV reservoir during ART.

Here we describe the development of ultrasensitive PCR-based assays to measure HIV persistence during ART. We optimized primers, probes, and reaction conditions to detect very low copy numbers of total HIV DNA, integrated HIV DNA, and 2-LTR circles in cell lysates, the use of which circumvents the labor-intensive step of DNA extraction. Of note, these assays can detect HIV sequences from the 6 major circulating subtypes (A, B, C, D, CRF01_A/E [A/E], and CRF01_A/G [A/G], which account for 85% of global HIV-1 variants [35]) and can be completed in a single day, making them well suited for large-cohort studies.

MATERIALS AND METHODS

Patient population.

Thirty-one HIV-seropositive individuals on stable suppressive ART and 15 chronically infected subjects with no history of ART enrolled in this study. All subjects signed informed consent forms approved by the Oregon Health and Science University and the Martin Memorial Health Systems review boards. None of the patients under ART experienced any detectable plasma viremia at the time of study, as assessed by viral load measurement using the Amplicor HIV-1 Monitor ultrasensitive method (Roche), which has a detection limit of 50 copies/ml of plasma (Table 1). These patients had had an undetectable viral load for a median of 6.5 years (range, 3.5 to 8.5 years). All patients underwent leukapheresis.

TABLE 1.

Subject characteristics^a

Characteristic	Untreated subjects	Subjects in which ART suppressed HIV
No. of subjects	15	31
Plasma HIV RNA load (no. of copies/ml)	16,000 (9,184–37,491)	<50 (<50–<50)
CD4+ T cell count (no. of cells/μl)	480 (334–508)	531 (431–874)
CD8+ T cell count (no. of cells/μl)	1206 (832–1,571)	780 (503–1,059)
CD4/CD8 ratio	0.34 (0.29–0.54)	0.75 (0.57–1.12)
Duration of HIV infection (yr)	5.6 (4.2–9.9)	7.7 (6.1–13.4)
Duration of ART (yr)	0 (0–0)	6.5 (3.5–8.5)

^aData represent medians (interquartile ranges).

Isolation of CD4⁺ T cells.

Peripheral blood mononuclear cells (PBMCs) were obtained from samples obtained by leukapheresis by Ficoll Hypaque density gradient centrifugation. Total CD4⁺ T cells were isolated from the PBMCs of successfully treated subjects and chronically infected donors using magnetic bead-based negative selection (Stem Cell Technologies). The purity of the enriched CD4⁺ T cells was generally greater than 95%, as assessed by flow cytometry.

Proteinase K digestion.

PBMCs, CD4⁺ T cells, and ACH-2 cells were centrifuged in a 1.5-ml microtube at 16,000 × g for 5 min, and the supernatants were carefully removed and discarded. Cell pellets were resuspended in a lysis buffer (10 mM Tris-HCl, pH 8.0, 50 nM KCl, 400 μg/ml proteinase K; Invitrogen) at the appropriate concentrations (20 × 10⁶ cells/ml for ACH-2 cells; 5 × 10⁶ to 10 × 10⁶ cells/ml for clinical samples) and digested for 12 to 16 h at 55°C in a heating shaker. Proteinase K was inactivated by heating the digested samples at 95°C for 5 min. Cell lysates were immediately used for HIV DNA quantification or stored at −20°C for future use.

Generation of quantification standards.

ACH-2 cells, which carry a single copy of the integrated HIV genome (36), were used to generate a standard curve for the total and integrated HIV DNA assays. Serial 10-fold dilutions of ACH-2 cells ranging from 3 × 10⁵ to 3 cells were amplified together with experimental samples and were used as standards for both HIV and CD3 gene quantifications.

As a standard for the 2-LTR circle assay, we cloned the human CD3 gene together with the 2-LTR circle junction obtained after in vitro infection of activated PBMCs with HIV-1 isolate LAI into a Topo TA cloning vector (Invitrogen). Serial 10-fold dilutions of the plasmid ranging from 6 × 10⁵ to 6 copies were amplified together with experimental samples and were used as standards for both 2-LTR and CD3 gene quantifications.

Quantification of total HIV DNA, integrated HIV DNA, and 2-LTR circles.

Fifteen microliters of the cell lysate was used in all preamplification reactions. In all PCRs, primers specific for the human CD3 gene (primers HCD3OUT5′ and HCD3OUT3′) were used to quantify the exact number of cells present in the reaction tube (Table 2). Preamplification of total HIV DNA and the CD3 gene was carried out in a 50-μl reaction mixture comprising 1× Taq polymerase buffer (Invitrogen), 3 mM MgCl₂, 300 μM deoxynucleoside triphosphates (Invitrogen), 300 nM each of the 4 primers (primers ULF1, UR1, HCD3OUT5′, and HCD3OUT3′), and 2.5 U Taq polymerase (Invitrogen). The first-round PCR cycle conditions were as follows: a denaturation step of 8 min at 95°C and 12 cycles of amplification (95°C for 1 min, 55°C for 40 s, 72°C for 1 min), followed by a final elongation step at 72°C for 15 min.

TABLE 2.

Sequences of primers and probes used in total HIV DNA, integrated HIV DNA, and 2-LTR circle assays

Target	Step	Primer or probe name	Sequencea (5′ to 3′)	Position in HXB2
Total HIV DNA	Preamplification	ULF1	ATG CCA CGT AAG CGA AAC TCT GGG TCT CTC TDG TTA GAC	452–471 (LTR 5′)
		UR1	CCA TCT CTC TCC TTC TAG C	775–793 (LTR-Gag)
	Real-time PCR	Lambda T	ATG CCA CGT AAG CGA AAC T	NAb
		UR2	CTG AGG GAT CTC TAG TTA CC	583–602 (LTR 5′)
		UHIV TaqMan	LC640-CAC TCA AGG CAA GCT TTA TTG AGG C-BBQ	522–546 (LTR 5′)
Integrated HIV DNA	Preamplification	ULF1	ATG CCA CGT AAG CGA AAC TCT GGG TCT CTC TDG TTA GAC	452–471 (LTR 5′)
		Alu1	TCC CAG CTA CTG GGG AGG CTG AGG	NA
		Alu2	GCC TCC CAA AGT GCT GGG ATT ACA G	NA
	Real-time PCR	Lambda T	ATG CCA CGT AAG CGA AAC T	NA
		UR2	CTG AGG GAT CTC TAG TTA CC	583–602 (LTR 5′)
		UHIV TaqMan	LC640-CAC TCA AGG CAA GCT TTA TTG AGG C-BBQ	522–546 (LTR 5′)
2-LTR circles	Preamplification	ULLTRF1	ATG CCA CGT AAG CGA AAC TCC TCA ATA AAG CTT GCC TTG A	9608–9628 (LTR 3′)
		ULTRR1	CTA ACM AGA GAG ACC CAG TAC	449–469 (LTR 5′)
	Real-time PCR	Lambda T	ATG CCA CGT AAG CGA AAC T	NA
		ULTRR2	GGT ACT AGC TTG AAG CAC CA	132–151 (LTR 5')
		U2LTR TaqMan	LC640-ACT CTG GTA ACT AGA GAT CCC TCA GAC C-BBQ	9663–9690 (LTR 3′)
CD3 gene	Preamplification	HCD3OUT5′	ACT GAC ATG GAA CAG GGG AAG	NA
		HCD3OUT3′	CCA GCT CTG AAG TAG GGA ACA TAT	NA
	Real-time PCR	HCD3IN5′	GGC TAT CAT TCT TCT TCA AGG T	NA
		HCD3IN3′	CCT CTC TTC AGC CAT TTA AGT A	NA
		CD3 TaqMan	LC640-AGC AGA GAA CAG TTA AGA GCC TCC AT-BBQ	NA

^aItalic sequences indicate the position of Lambda T in ULF1. LC640, LC Red 640; BBQ, BlackBerry quencher.

^bNA, not applicable.

Integrated HIV genomes were amplified by using the same mix used for total HIV DNA, with the exception that the reverse primer (UR1) was replaced by the Alu1 and Alu2 primers (300 nM each) and the concentration of ULF1 was reduced (150 nM). Given the high number of Alu elements within the human genome, abundant amplifications of inter-Alu sequences occurred simultaneously with the amplification of Alu LTR sequences. To remain in the exponential phase, only 12 cycles of amplification were performed. The PCR cycle conditions were as follows: a denaturation step of 8 min at 95°C and 12 cycles of amplification (95°C for 1 min, 55°C for 1 min, 72°C for 10 min), followed by an elongation step of 15 min at 72°C.

2-LTR circle sequences were amplified with a mix containing the CD3-specific primers together with the ULLTRF1 and ULTRR1 primers (all at 300 nM). The PCR conditions were as follows: a denaturation step of 8 min at 95°C and 16 cycles of amplification (95°C for 30 s, 55°C for 30 s, 72°C for 1 min), followed by an elongation step of 15 min at 72°C. All preamplifications were carried out on a Mastercycler pro-S instrument (Eppendorf).

The second rounds of PCR were carried out in real time on a Rotor-Gene Q instrument (Qiagen) with the Rotor-Gene probe master mix (Qiagen) following the manufacturer's instructions. All reactions (for total HIV DNA, integrated HIV DNA, 2-LTR circles, and CD3) were performed in a final volume of 20 μl containing 6.4 μl of a 1/10 dilution of the first PCR products. The appropriate sets of primers (1,250 nM Lambda T and UR2 for total and integrated HIV DNA, Lambda T and ULTRR2 for 2-LTR circles, and HCD3IN5′ and HCD3IN5′ for the CD3 gene) were added to the Rotor-Gene probe master mix. The UHIV TaqMan probe (200 nM) was added to the total and integrated HIV DNA reaction mixtures, whereas the same concentration of the U2LTR TaqMan probe was used for the 2-LTR reaction. For CD3 quantification, 200 nM the CD3 TaqMan probe was used. The same amplification steps were used for all reactions: a denaturation step (95°C for 4 min), followed by 40 cycles of amplification (95°C for 3 s, 60°C for 10 s).

In vitro infection and cloning.

PBMCs (4 × 10⁶) were stimulated for 3 days with 1 μg/ml phytohemagglutinin (PHA) and 10 ng/ml interleukin-2 (IL-2) and infected with HIV isolates (see Table S1 in the supplemental material) in a 15-ml Falcon tube for 4 h at 37°C in the presence of 200 nM raltegravir when appropriate. Cells were washed twice in complete medium (RPMI 1640, 10% fetal bovine serum) and resuspended in culture medium supplemented with 10 ng/ml IL-2. After 24 h, 48 h, and 72 h, 1 × 10⁶ cells were collected, washed, and pelleted by centrifugation at 16,000 × g for 5 min. Dry pellets were conserved at −80°C until HIV DNA quantification.

Similar infections were carried out to produce a maximal amount of the three HIV molecular forms for subsequent cloning to generate standards for all HIV clades. In these experiments, freshly PHA- and IL-2-stimulated PBMCs were added after 3 days and 6 days to increase HIV replication. After 9 days of culture, infected cells were pelleted and used to amplify the LTR-gag sequence (total HIV DNA) and the 2-LTR junction (2-LTR circle assay) using primers annealing outside the region of interest. Amplified products were cloned in a Topo TA cloning vector (Invitrogen).

RESULTS

Principle of the assays.

To precisely measure the frequency of cells harboring total HIV DNA, integrated HIV DNA, and 2-LTR circles, we developed highly sensitive real-time nested PCR-based assays which allow the detection of very low HIV DNA copy numbers in 100,000 cells. For each assay, the cellular input was precisely quantified by measuring the number of human genome equivalents (CD3 gene, 2 copies per cell) in the same reaction. In a first round of PCR, HIV DNA (total or integrated HIV DNA or 2-LTR circles) and the CD3 gene (2 copies per cell) were coamplified in a single well. The products of this first PCR were amplified in a second round of PCR using inner primers and probes annealing in the appropriate regions (Fig. 1A).

FIG 1.

(A) Principle of the PCR-based assays used to measure total HIV DNA, integrated HIV DNA, and 2-LTR circles in samples from HIV-infected individuals. Cells are digested overnight (O/N) in proteinase K (PK), and the cell lysates are used in a pre-PCR step in which one of the three molecular forms is preamplified together with the CD3 gene. Preamplified products are diluted and used in a second amplification reaction (HIV or CD3) in which each form is quantified by real-time PCR. The frequency of cells harboring each molecular form is calculated from the ratio of the HIV copy number/2 × (CD3 copy number). (B) Positions of the primers and probes used in the 3 assays.

To preamplify total HIV DNA, we used oligonucleotides specific for the LTR (U3-R junction, primer ULF1) and gag (primer UR1) regions. Integrated HIV DNA was preamplified by using the same forward primer (primer ULF1, which also anneals in the 3′ LTR) together with two oligonucleotides specific for human Alu sequences. As previously described (24, 37), the use of two Alu-specific primers optimizes the probability of amplifying an LTR sequence, as Alu elements are present in either orientation relative to the integrated provirus (Fig. 1B). The 2-LTR junction was preamplified by using a forward oligonucleotide annealing in the LTR R region (primer ULLTRF1) together with a reverse primer specific for the LTR U3 region (primer ULTRR1). Both forward primers used in these first amplification steps (primer ULF1 for total and integrated HIV DNA, primer ULTRF1 for 2-LTR circles) were extended with a lambda phage-specific heel sequence at the 5′ end of the oligonucleotide. Using a lambda phage-specific primer (Lambda T) in the second round of amplification, only products from the first-round PCR could be amplified. The same primers and probes were used for the second-round amplifications in the total HIV DNA and integrated HIV DNA assay: the Lambda T primer was used together with an oligonucleotide specific for the LTR U5 region (primer UR1) and with an hydrolysis probe that anneals at the end of the LTR R region (probe UHIV TaqMan). 2-LTR junctions were also further amplified in a second round of PCR using the Lambda T primer together with an oligonucleotide specific for the LTR U3 region (primer ULTRR2) and a probe annealing at the end of the LTR U5 region (probe U2LTR TaqMan).

For the total and integrated HIV DNA assays, serial dilutions of ACH-2 cells, which carry a single copy of the integrated HIV genome, were used to create a standard curve. As a standard for the 2-LTR circle assay, we used serial dilutions of a plasmid in which the 2-LTR circle junction amplified following in vitro infection with HIV-1 isolate LAI was cloned together with the human CD3 gene. The ratio between the number of copies of a given form (total HIV DNA, integrated HIV DNA, and 2-LTR circles) and the number of cells used in the reaction (number of CD3 copies) was calculated to obtain the frequency of cells harboring each of these forms.

Design of universal primers and probes.

To design a set of universal primers and probes that would allow the quantification of total HIV DNA, integrated HIV DNA, and 2-LTR circles from the A, B, C, D, CRF01_A/E, and CRF02_A/G HIV-1 subtypes, we first aligned all subtype reference sequences from these clades. We identified regions conserved among all subtypes in the LTR-gag region (for total and integrated DNA) and at the junction of the 5′ LTR and 3′ LTR (for 2-LTR circles) and designed primers and probes matching these regions (Table 2). Degenerate bases were introduced in primers ULF1 (the forward oligonucleotide used in the first amplification for total and integrated DNA assays) and ULTRR1 (the reverse oligonucleotide used in the first amplification for the 2-LTR circle assay). Sequences of these primers and probes were further aligned with all sequences from the HIV database, and the percentages of sequences from each clade that displayed 0 or 1 and more than 1 mismatch were counted (Table 3). For the total and integrated HIV DNA assays, 2,202 to 3,939 individual LTR-gag sequences were examined. The analysis revealed that the oligonucleotides ULF1, UR1, and UR2 and the probe UHIV TaqMan annealed to 91%, 98%, 96%, and 98% of all HIV-1 published sequences with 1 mismatch or less, respectively. Similarly, the primers and probes used in the 2-LTR circle assays annealed to the vast majority of the published HIV-1 sequences (314 to 2,228 sequences were examined) with less than 2 mismatches (96%, 93%, 70%, and 94% for oligonucleotides ULLTRF1, ULTRR1, and ULTRR2 and the probe U2LTR TaqMan, respectively). The relatively low frequency of sequences annealing to ULTRR2 with 1 mismatch or less (70%) was mostly attributed to subtype D sequences (3.5%). However, the majority of the sequences from the D clade (96.5%) annealed to ULTRR2 with only 2 mismatches, indicating that this oligonucleotide could also be used to quantify 2-LTR circles from this clade. Altogether, the results from this in silico analysis indicate that the oligonucleotides and probes that we designed recognize the vast majority of the circulating HIV-1 clades with 1 mismatch or less.

TABLE 3.

Percent identity of the primers and probes used in the three assays compared to sequences from HIV-1 circulating clades^a

Clade	ULF1			UR1			UR2			UHIV TaqMan			ULLTRF1			ULTRR1			ULTRR2			U2LTR TaqMan
	N sequences	% mismatch		N sequences	% mismatch		N sequences	% mismatch		N sequences	% mismatch		N sequences	% mismatch		N sequences	% mismatch		N sequences	% mismatch		N sequences	% mismatch
		0–1	>1		0–1	>1		0–1	>1		0–1	>1		0–1	>1		0–1	>1		0–1	>1		0–1	>1
A	108	96	4	69	97	3	90	94	6	90	98	2	4	100		108	94	6	109	85	15	3	100	0
B	1,445	91	9	2,470	98	2	2,250	91	9	1,410	98	2	639	96	4	1,437	96	4	841	80	20	150	97	3
C	423	93	7	682	99	1	719	98	2	443	97	3	222	96	4	413	85	15	514	55	45	132	90	10
D	59	96	4	38	95	5	55	98	2	61	98	2	14	100		57	83	17	57	3.5	96.5	6	100	0
A/E	96	91	9	436	100	0	243	99	1	82	100	0	61	100		97	90	10	96	87	13	21	91	9
A/G	117	82	18	244	99	1	251	100		116	98	2	18	100		116	98	2	34	85	15	2	100	0
All clades	2,248	91	9	3,939	98	2	3,608	96	4	2,202	98	2	958	96	4	2,228	93	7	1,651	70	30	314	94	6

^aN sequences, number of sequences analyzed; % mismatch, percentage of analyzed sequences that annealed to the indicated primer or probe with 0 (identity) or 1 mismatch or more than 1 mismatch.

Linearity and sensitivity of the assays.

To assess whether these universal primers and probes could be used to precisely quantify low copy numbers of HIV DNA from the A, B, C, D, and A/E clades by real-time PCR, we infected stimulated PBMCs with primary isolates from each clade (2 isolates per clade; see Table S1 in the supplemental material). We successfully amplified and cloned LTR-gag regions from the 10 HIV isolates from these 5 different subtypes. Sequencing of the cloned sequences confirmed that the primers and probes described above were annealing to well-conserved regions of LTR and gag (see Fig. S1 in the supplemental material). Serial dilutions of these constructs ranging from 300,000 copies to 3 copies were used in the total HIV DNA assay. Serial dilutions of ACH-2 cells, which contain a single copy of the HIV LAI genome per cell, were used as controls. The linear regression obtained from amplification of serial dilutions of LTR-gag junctions was linear over a 6-log₁₀-unit range (Fig. 2A), and the assay allowed the detection of 3 copies of HIV DNA from all clades (Fig. 2B). Importantly, the linearity of the amplification was conserved among all clades and similar to the one obtained with serial dilutions of ACH-2 cells.

FIG 2.

(A) Representative curves obtained after amplification of serial dilutions (300,000 copies to 3 copies) of HIV DNA from ACH-2 cells (solid bold lines) and the LTR-gag region cloned from the A08483M1 (D subtype; solid thin lines) and NI1149 (CRF01_A/E subtype; dashed lines) isolates. Norm. Fluoro., normalized fluorescence. (B) C_T values generated after amplification of the LTR-gag region (total HIV DNA assay) from 10 HIV-1 isolates from 5 different subtypes. ACH-2 cells were used as a standard. (C) C_T values generated after amplification of the 2-LTR junction (2-LTR circle assay) from 5 HIV-1 isolates from 3 different subtypes. Serial dilutions of a plasmid containing the 2-LTR junction generated following in vitro infection with the LAI isolate were used as a standard.

We used a similar strategy to assess the linearity and sensitivity or the 2-LTR circle assay. 2-LTR junctions generated after infection with 7 isolates (1 subtype A, 2 subtype B, 2 subtype C, and 2 subtype A/E isolates) were successfully amplified and cloned. Sequencing confirmed that the primers and probes designed for this assay annealed to well-conserved regions among all the tested clades (see Fig. S2 in the supplemental material). Serial dilutions of 5 of these constructs ranging from 600,000 copies to 6 copies were used in the 2-LTR circle assay. Similar to the total HIV DNA quantification, the linear regression obtained from amplification of serial dilutions of the 2-LTR junction was linear over a 6-log₁₀-unit range, and the assay allowed the detection of 6 copies of HIV DNA from all clades (Fig. 2C).

To demonstrate the robustness and reproducibility of our assays, we analyzed the standard curves of the integrated HIV DNA assay generated by 4 independent operators over a 12-month period (see Fig. S3 in the supplemental material). The threshold cycle (C_T) values for the HIV and CD3 gene reactions (see Fig. S3A and B, respectively, in the supplemental material) were highly consistent between the different operators. As a result, the slopes of the standard curves were very reproducible, with efficiencies constantly being >85% for both HIV and CD3 amplifications (see Fig. S3C in the supplemental material). This demonstrated that our assays to measure HIV DNA forms are robust and reproducible.

To determine if the number of input cells would affect the sensitivity and linearity of our assay, we serially diluted known numbers of plasmids containing a single copy of the gag-LTR region (range, 2 to 2,000 copies) in serial dilutions of uninfected PBMCs (range, 10,000 to 300,000 cells). We observed that HIV DNA molecules were detected with similar efficacy when the number of input cells ranged from 10,000 to 300,000 (see Fig. S4A in the supplemental material). Conversely, the quantification of the CD3 gene copy numbers was not affected by the number of HIV DNA copies present in the sample (see Fig. S4B in the supplemental material).

We also tested for the presence of PCR inhibitors, which may alter the sensitivity of the assay, in the proteinase K lysates. We diluted samples from 2 HIV-infected subjects in serial 2-fold dilutions in proteinase K lysing buffer and measured CD3 and HIV copy numbers. The number of copies detected in each well reflected the dilution factor for both CD3 and HIV (see Fig. S5A and B, respectively, in the supplemental material). Therefore, the ratio (HIV copy number/CD3 copy number) was not altered when the lysates were used pure or at 1/2, 1/4, and 1/8, demonstrating the absence of PCR inhibitors in our cell lysates (see Fig. S5C in the supplemental material).

It is well established that background amplification can occur in Alu-based PCR assays. Several studies have shown that the sense oligonucleotide can prime the formation of a single-stranded DNA from all LTR-containing HIV DNA, leading to an overestimation of the actual integrated HIV-1 DNA copy number (24, 27). To assess the contribution of this phenomenon, we performed our assay with or without Alu-specific primers in the first-round PCR using 4 samples from ART-naive subjects and 4 samples from virally suppressed subjects. The number of HIV copies detected in the absence of Alu-specific primers was low and negligible compared to the number of HIV copies detected when the Alu-specific primers were present in the PCR (see Fig. S6A in the supplemental material). Of note, the measurement of the CD3 gene was not affected by the presence or absence of the Alu-specific primers (see Fig. S6B in the supplemental material). We then calculated the ratio of the HIV copies/CD3 copies to obtain the frequency of cells with integrated HIV DNA. On average, the frequency of cells carrying integrated HIV DNA was 47-fold times higher in the presence of the Alu-specific primers (range, 17- to 164-fold difference; see Fig. S6C in the supplemental material), demonstrating that the unspecific signal detected in the absence of Alu-specific primers is negligible.

Detection of total HIV DNA, integrated HIV DNA, and 2-LTR circles following in vitro infection with primary isolates from different HIV clades.

To follow the synthesis and fate of total HIV DNA, integrated HIV DNA, and 2-LTR circles using our assays, we infected prestimulated PBMCs with isolates from the A, B, C, D, and A/E clades and measured these 3 forms after 1, 2, and 3 days of culture (Fig. 3). As expected, the frequency of cells harboring total HIV DNA largely exceeded that of cells harboring integrated DNA and 2-LTR circles. Infection with the 5 clades of isolates showed readily detectable levels of total HIV DNA at 24 h postinfection, whereas integrated DNA and 2-LTR circles were barely detectable at this early time point. Integrated HIV DNA and 2-LTR circles were detected in all cultures at 48 and 72 h postinfection.

FIG 3.

Kinetics of production of total HIV DNA, integrated HIV DNA, and 2-LTR circles following in vitro infection of prestimulated PBMCs with HIV-1 isolates from 5 different clades. (A) Subtype A isolate KNH1144; (B) subtype B isolate US4; (C) subtype C isolate SE364; (D) subtype D isolate A08483M1; (E) subtype CRF01_A/E isolate CM235.

To further demonstrate the specificity of these assays, we repeated these experiments in the presence of the integrase inhibitor raltegravir (Fig. 4). Low levels of total HIV DNA were transiently detected following infection with the CRF01_A/E clade isolate CM235 at 24 h in the presence of raltegravir (Fig. 4A), whereas integrated DNA was undetectable at all time points in the presence of the drug (Fig. 4B). Interestingly, a larger number of 2-LTR circles was generated when raltegravir was present in the culture, and this number slowly decreased with time (Fig. 4C). This accumulation of 2-LTR circles upon raltegravir treatment is in line with previous in vitro and in vivo observations (8, 9, 38). We repeated these experiments with HIV-1 isolates from the A, B, C, D, and A/E clades and observed similar trends: the presence of raltegravir in the cultures of infected cells led to lower numbers of total HIV DNA molecules (Fig. 4D), low to undetectable levels of integrated HIV DNA (Fig. 4E), and increased numbers of 2-LTR circles (Fig. 4F). Altogether, these experiments demonstrated the specificity of our assays and their capacity to detect HIV DNA molecular forms generated following infection with HIV-1 isolates from 5 subtypes.

FIG 4.

(A to C) Kinetics of production of total HIV DNA (A), integrated HIV DNA (B), and 2-LTR circles (C) following in vitro infection of prestimulated PBMCs with HIV-1 subtype CRF01_A/E isolate CM235 in the presence or absence of the integrase inhibitor raltegravir. Values are means ± standard deviations. (D to F) The frequencies of cells harboring total HIV DNA (D), integrated HIV DNA (E), and 2-LTR circles (F) were measured at 24 h after infection of prestimulated PBMCs with HIV-1 isolates from 5 different subtypes in the presence or absence of raltegravir. Values are means ± standard deviations.

Detection of total HIV DNA, integrated HIV DNA, and 2-LTR circles in CD4⁺ T cells from HIV-infected subjects.

We used these assays to measure the frequency of CD4⁺ T cells harboring total DNA, integrated HIV DNA, and 2-LTR circles in 15 ART-naive subjects and in 31 individuals who had received ART for at least 3 years (Fig. 5). The three forms (total HIV DNA, integrated HIV DNA, and 2-LTR circles) were readily detected in samples from all viremic subjects (Fig. 5A). Total HIV DNA, integrated HIV DNA, and 2-LTR circles were detected in 100%, 94%, and 77% of samples from virally suppressed subjects, respectively.

FIG 5.

(A) Proportion of samples obtained from virally suppressed individuals (subjects receiving ART [ART]) and untreated HIV-infected subjects (viremic subjects [VIR]) with detectable levels of total HIV DNA, integrated HIV DNA, and 2-LTR circles. The numbers of samples tested from each group and in each assay are indicated. (B) Quantifications of total HIV DNA, 2-LTR circles, integrated HIV DNA, and plasma HIV RNA in samples obtained from virally suppressed individuals (blue symbols) and untreated HIV-infected subjects (red symbols). Error bars are medians and interquartile ranges. Undetectable values for 2-LTR circles and integrated HIV DNA are depicted by empty symbols. Dotted line, limit of detection. (C) Ratio of total HIV DNA/integrated HIV DNA in CD4⁺ T cells obtained from virally suppressed individuals.

All subjects on ART had an undetectable viral load using conventional assays (limit of detection = 50 copies per ml of plasma), whereas the median viral load in untreated individuals was 16,000 HIV-1 RNA copies/ml of plasma (P < 0.0001; Fig. 5B). The levels of total HIV DNA were significantly higher in CD4⁺ T cells from viremic subjects than those from individuals on ART (2,244 and 976 copies/10⁶ CD4⁺ T cells, respectively; P = 0.0003). Similarly, 2-LTR circles, which have been proposed to be a surrogate for active HIV replication (39), were detected at higher levels in untreated HIV-infected subjects than virally suppressed individuals (135 and 23 copies/10⁶ CD4⁺ T cells, respectively; P = 0.0004). In contrast, although there was a trend for higher levels of integrated HIV DNA in untreated subjects, the difference between the two groups of individuals did not reach statistical significance (837 and 338 copies/10⁶ CD4⁺ T cells, respectively; P = 0.23), indicating that ART minimally affects the size of the pool of cells carrying integrated viral genomes.

We determined the coefficient of variation associated with each assay by repeating the quantifications of total and integrated HIV DNA as well as 2-LTR circles in 6 samples from this cohort (see Fig. S3D in the supplemental material). All procedures were repeated, from thawing of cryopreserved PBMCs to counting, digestion, preamplification, and real-time PCR, by 2 different operators 6 months apart. The coefficients of variation were found to be low (0.26, 0.43, and 0.53 for total HIV DNA, integrated HIV DNA, and 2-LTR circles, respectively), demonstrating the robustness of these assays.

We calculated the ratio between total and integrated HIV DNA in both groups of subjects (Fig. 5C). Compared to virally suppressed individuals, viremic donors displayed significantly higher total HIV DNA/integrated HIV DNA ratios (means = 2.6 and 6.7, respectively; P = 0.0009), indicating that nonintegrated DNA largely predominates in subjects with untreated HIV infection but not in virally suppressed subjects. Altogether, our results indicate that by using these highly sensitive assays that do not require nucleic acid extraction, at least one of the three markers of HIV persistence can be detected in CD4⁺ T cells from all HIV-infected subjects and demonstrate that the proportion of virally suppressed subjects harboring 2-LTR circles was previously underestimated.

DISCUSSION

Since the discovery of viral reservoirs in HIV-infected subjects receiving suppressive ART, measuring the degree of viral persistence has been one of the greatest challenges in the field of HIV research (17). Latently infected CD4⁺ T cells are extremely rare, and even the most sensitive assays have a detection limit that is often driven by the amount of available biological material: the frequency of latently infected cells harboring replication-competent virus in subjects who have received ART for several years is about 1 cell in a million resting CD4⁺ cells (3–5, 16). Recent findings indicate that this extremely low frequency, which was originally measured by using the quantitative viral outgrowth assay, may have substantially underestimated the size of the latent reservoir, which may be up to 60-fold greater than was previously thought (21). Therefore, alternative assays that provide a better estimate of the size of the latent reservoir are needed.

Measuring the frequency of cells harboring HIV DNA (integrated or not) using PCR-based methods has often been criticized, as only a small proportion of CD4⁺ T cell-associated HIV DNA encodes replication-competent virus (20). However, these methods present several important advantages: PCR-based assays performed on ex vivo clinical samples are more robust and precise than culture-based assays, can be performed on large numbers of samples simultaneously, and, given their sensitivity, require smaller numbers of cells for analysis (18).

The majority of PCR-based methods that have been developed so far allow the quantification of three HIV DNA forms (total HIV DNA, integrated HIV DNA, and 2-LTR circles) from clade B viruses, which are predominant in Europe, North America, and Australia, where most of the cohorts of HIV-infected subjects were originally established (24, 27, 28). As large studies aimed at testing the efficacy of vaccines, antiretroviral combinations, and eradication strategies have been initiated in other regions of the world, there is an urgent need to establish assays that can detect and quantify HIV DNA from other subtypes from group M represented worldwide, such as subtypes A (East Africa), C (southern Africa, India, Nepal), D (eastern and Central Africa), CRF01_A/E (Thailand), and CRF02_A/G (West Africa and central Europe). The three sensitive nested real-time PCR assays that we developed allow the quantification of total HIV DNA, integrated HIV DNA, and 2-LTR circles from subtypes A, B, C, D, and CRF01_A/E. The three assays are highly sensitive and robust and can be performed in a single day without the need for nucleic acid extraction.

ACH-2 cells, which were used for the generation of standard curves in the total and integrated HIV DNA assays, are not entirely transcriptionally silent, and a minor fraction of these cells produces viral particles, which may result in more than 1 HIV DNA molecule per cell. This warrants a careful quantification of the DNA copy number in a given lot of ACH-2 cells before using cells of the lot as a standard in these assays. As presented in Fig. 2B, serial dilutions of ACH-2 cells and serial dilutions of plasmids containing similar and known copy numbers of HIV DNA molecules gave similar amplification curves, indicating that the ACH-2 cells that we used in this study contained ≈1 copy of HIV DNA per cell.

Using our integrated HIV DNA assay, we estimated that the frequency of CD4⁺ T cells harboring integrated HIV DNA is close to 300 in a million cells. Of note, this frequency is 5 times higher than the frequency of cells with replication-competent virus measured by Ho et al. (21). Our integrated HIV DNA assay may provide a more sensitive estimate of the size of the reservoir than the viral outgrowth assay, which may underestimate this frequency by 60-fold.

Whether ART completely blocks de novo infection or whether low levels of residual HIV replication persist in virally suppressed subjects is still unclear (6, 8, 11). Residual levels of viremia can be measured by using a single-copy assay (7), but this assay is difficult to implement in large-cohort studies, and the clinical relevance of these traces of virus remains largely unknown (20). Alternatively, the presence of 2-LTR circles, a vestige of the viral genome that is generated during abortive infection events, has been proposed to be a surrogate marker for ongoing viral replication during ART (39, 40). However, the use of HIV episomes as markers of active viral replication has been challenged, as they may be much more stable than was originally thought (41, 42). Using our highly sensitive assay for 2-LTR circles, we observed that the majority (77%) of subjects on ART displayed detectable levels of episomes in CD4⁺ T cells. To our knowledge, this proportion of virally suppressed subjects displaying detectable levels of 2-LTR circles is the highest reported so far, demonstrating the highly sensitive nature of our assay. This high proportion of virally suppressed subjects showing detectable levels of 2-LTR circles may be attributed to one of the unique features of our assay, i.e., the lack of the need for DNA extraction. Indeed, small DNA molecules, including episomes, may be lost during the DNA extraction process (43). Our assay is performed on cell lysates, which circumvents this potential issue. The presence of 2-LTR circles as direct evidence of ongoing viral replication during suppressive ART has been challenged (41, 42). However, there is no doubt that the increase in the frequency of cells harboring these episomes upon intensification of therapy by use of the integrase inhibitor raltegravir results from the inhibition of cryptic viral replication during ART (8, 9). Our experiments conducted in vitro indicate that our method has the ability to detect this increase in episomes upon raltegravir treatment in subjects infected with all HIV subtypes. Therefore, this method would be well suited to determine if a particular HIV subtype has a greater ability to undergo residual levels of replication during suppressive ART, by intensifying therapy with an integrase inhibitor in populations infected with different clades and monitoring for an increase in the levels of 2-LTR circles upon administration of the drug, as previously reported in populations infected with clade B viruses (8, 9).

Because large deletions and mutations are often present in 2-LTR circles, precise quantification of these episomes is challenging (38, 44). We cannot exclude the possibility that our assay underestimates the frequency of cells harboring 2-LTR circles as a result of potential large deletions within the regions to which our sets of primers and probes anneal.

In conclusion, we report the development and validation of ultrasensitive assays to measure HIV persistence in HIV-infected individuals from multiple geographical regions. These assays are relatively inexpensive, do not require DNA extraction, and can be completed in a single day. Therefore, they are perfectly adapted to monitor HIV persistence in large cohorts of HIV-infected individuals, and given their sensitivity, they can be used to monitor the efficacy of therapeutic strategies aimed at interfering with HIV persistence after prolonged ART.